S. Afr. 1. Bot., 1987,53(2): 173 -176 173

Chemosystematic studies of the genus Combretum (Combretaceae). I. A convenient method of identifying species of this genus by a comparison of the polar constituents extracted from leaf material

J.D. Carr and C.B. Rogers* Sandton, Transvaal and Department of Chemistry, University of Durban-Westville, Private Bag X54001, Durban, 4000 Republic of South Africa

Accepted 12 December 1986

It is now generally accepted that chemical data are becoming increasingly important as an adjunct to morphological evidence in the classification of species and for resolving taxonomic problems. We describe a chemical method that can be used for the rapid identification of species of the genus Combretum Loefl. in southern Africa and for resolving certain classification problems in this genus. Results of the analysis of 26 species are given.

Bykomstig tot morfologiese kenmerke is chemiese inligting 'n nuttige hulpmiddel by die klassifikasie van spesies en die oplos van taksonomiese probleme. 'n Chemiese metode met behulp waarvan die spesies van die genus Combretum Loefl. vinnig ge"identifiseer kan word en sekere klassifikasieprobleme in die genus opgelos kan word, word beskryf. Resultate van die analise van 26 suider-Afrikaanse spesies word verskaf.

Keywords: Combretaceae, Combretum, chemical analysis, chemotaxonomy

'To whom correspondence should be addressed

Introduction extractives serve as a 'finger-print' for a particular species and The isolation and identification of the novel la-hydroxy could thus be useful for identification and classificatory cycloartenoids mollic acid (1) and jessic acid (2) (Figure 1) purposes (Figure 2). Similar studies using flavonoids as plus their xylosides, arabinosides and glucosides in Cam­ markers have been used by Harbome et al. (1975) to resolve bretum malle R. Br. and C. elaeagnaides Klotzsch leaves classification problems in a number of different families. (Osborne & Pegel 1984; Pegel & Rogers 1985; Rogers & Thevan 1986) prompted a comprehensive chemical inves­ Materials and Methods tigation of other Cambretum species. The resultant chemo­ An initial survey of leaf material from 17 species collected taxonomic survey carried out over several years on leaf from the National Botanic Gardens, Harare, Zimbabwe was material from 26 southern African species (Table 1) has conducted in 1977 . Air-dried, milled leaf material (70 - 100 g) yielded some interesting results of chemical and taxonomic from each species was extracted (Soxhlet) successively with significance. light petroleum, ether and acetone (8 h each), and the extracts Many of the species contain complex mixtures of tri­ evaporated to dryness under reduced pressure. A sample from terpenoids and other secondary metabolites whereas others each extract [100 mg rnl - 1 CHCb : ethanol (1: 1, V IV)1 was yielded relatively simple extracts. It was found that the polar analysed by thin layer chromatography (TLC) on Merck silica and therefore generally more complex extractives from leaf gel GF254 and the chromatograms recorded photographically material exhibited a unique chemical profile of compounds (Polaroid SX-70 Land camera Model 2; close-up lens 121) for each species. It has become apparent that these polar (Figure 2). Solutions were applied to the plates in bands 1 cm wide (2 applications) and allowed to develop by ascending chromatography for 9 cm. The polar secondary metabolites of interest were found in the ether and acetone extracts, and were best separated on TLC by the solvent system CHCI: ethyl acetate: formic acid (5 :4: 1, V IV). For spot detection a spray reagent consisting of anisaldehyde (5 rnI), conc. H2S04 (5 m!) and ethanol (90 rnI) was found to be most versatile, the coloured bands forming after the sprayed plates had been heated at 110° for 2 - 5 min. Because it appears to be ubiquitous in all leaf material, the phytosterolin, sitosteryl-B­ R (I) H D-glucoside, was applied to each plate to act as a marker to -D-glueosyl which Rf values could be related. -D-xylosyl Recently this method was modified when it was found that -L-arabinosyl digesting fresh leaves (± 5 g) in methanol (75 - 100 rnI) overnight at room temperature yielded an extract containing all the polar compounds found in the ether and acetone extracts. The methanolic solution decanted from the leaves (2) H is concentrated under vacuum and the water extracted from -L-arabinosyl the fresh leaves by the methanol is removed from the final extract by azeotropic distillation with benzene (2 x 20 ml). Figure 1 The I-a-hydroxyeycloartenoids mollie aeid (1) and jessie TLC analysis of this extract is then carried out as described aeid (2). above. 174 S.-Afr. Tydskr. Plantk., 1987, 53(2)

Table 1 List of Combretum species analysed; To assess the extent geographic and environmental factors relative abundance of triterpenoid and flavonoid­ might have in causing intraspecific variation, the polar type compounds indicated by crosses constituents in the leaves from C. molle specimens collected from the localities shown in Figure 3, were compared with TLC evidence for respect to their mollie acid and mollie acid xyloside, ara­ Section/ species Triterpenoids Flavonoids binoside and glucoside contents. This species is an ideal model to use, because of its wide distribution in southern Africa and Ciliatipetala C. molle Loefl. xxxxx because it is regarded as an aggregate species (Exell 1978). C. edwardsii Exell xxxxx XX In addition, leaves from trees growing in the National Botanic C. apiculatum Sond. subsp. leutweinii Garden, Harare cultivated from seeds collected in Angola, (Schinz) Exell X Zambia and Malawi were also examined. This effectively C. moggii Exell XX XX extended the distribution of specimens analysed to include the C. petrophilum Retief XX XX central African region. Angustimarginata Despite obvious differences in the appearance of leaves C. caffrum (Eckl. & Zeyh.) Kuntze X from coastal specimens (Durban area-large, glabrescent) C. kraussii Hochst xxxx XX compared with those from inland (Transvaal and Zimbabwe­ C. nelsonii Duemmer XX X smaller, densely tomentose), all specimens with the exception C. woodii Duemmer xxxx of that from Punda Milia showed the same four triterpenoid C. vendae Van Wyk XX X compounds to be present; the Punda Milia specimen provided H ypocrateropsis the typical C. molle pattern plus an additional, apparently C. celastroides Welw. ex Laws. subsp. closely related triterpenoid glycoside that has still to be celastroides X X identified. Furthermore seasonal influences could be dis­ C. imberbe Wawra xxxxxx X counted as leaves collected in Durban in January, Malelane C. padoides Engl. & Diebs. xxxx in February, Louwsberg Natal in May and Zimbabwe in July Conniventia gave the same pattern of extracts, although quantitative C. microphyllum Klotzsch X X differences which could not be detected by TLC may have C. paniculatum Vent. X X been present. Poivrea To complement this study, specimens from the following C. bracteosum (Hochst.) Brandis three species were also collected from more than one locality: C. mossambicense (Klotzsch) Engl. XX XXX C. hereroense Schinz: Kadoma, Zimbabwe; Louwsberg Lasiopetala Nature Reserve, Natal and Durban - seeds from Matopos C. obovatum F. Hoffm. XX National Park, Zimbabwe; Elaeagnoida C. imberbe Wawra: Kadoma, Zimbabwe; Hazyview, eastern C. elaeagnoides Klotzsch xxxx XX Transvaal and Durban - seeds from Botswana; Breviramea C. paniculatum Vent.: Harare, Zimbabwe, Sandton, Trans­ C. hereroense Schinz X XXX vaal and Durban - seeds from Harare. Spathulipetala In each case there was no discernible difference in the C. zeyheri Sond. XXX XX extracts obtained. The invariability of the C. hereroense Glabripetala extracts are particularly significant since the distinguishing C. fragrans F. Hoffm. XXX compounds in the methanol extract are not triterpenoids but Megalantherum flavonoids of which the major one has been isolated and C. wattii Exell X identified as ( + )-catachin. Since this compound and those Magrostigmatea related to it in the extract give characteristic bright orange C. sp. nov. (proposed XXX XX colours with the TLC spray reagent, they serve as ideal C. mkuzensis) markers for this species. Based on the colours formed on the 'TLC plates in response to the spray reagent, a tentative identification of flavonoids (orange-yellow) and triterpenoids (pink-blue) in all species studied is given in Table 1. However, The simplicity of this modified procedure and the small until all the compounds shown to be present by TLC are quantity of leaf material that is required are distinct advantages isolated and identified, the results in Table 1 can be considered and extractions could be carried out in the field. In the most only as a guide. recent study 26 species of the genus Combretum were analysed by this method (Table 1). Results of taxonomic significance Results and Discussion 1. C. nelsonii Duemmer, C. woodii Duemmer, and C. For this method to be reliable as a source of chemotaxonomic kraussii Hochst. evidence, the characterizing extractives should be independent The chemical profiles obtained for C. nelsonii and C. woodii of geographical, seasonal and other environmental influences. in this study do not support their inclusion in the synonymy Lower terpenoids such as the monoterpenoids can be sensitive of C. kraussii as indicated by Exell (1970, 1978). All three to all three influences (Loomis 1967), but this does not seem give methanol extracts that are quite different by TLC; C. to be the case with the more complex triterpenoids and their woodii contains predominantly triterpenoid-type compounds glycosides in this particular genus of the Combretaceae. This with high Rf values whereas in C. kraussii the majority of is possibly due to the greater complexity of triterpenoid compounds have lower Rf values. Compared with these two, biosynthesis, considered to be an extremely ancient character C. nelsonii gives a relatively simple extract which is composed (Smith 1976). of predominantly flavonoid-type compounds. These findings S. Afr. J. Bot., 1987, 53(2) 175

A

B c D

Figure 2 TLC chromatograms of methanol extracts from: I. C. molle; 2. C. edwardsii; 3. C. woodii; 4. C. padoides; 5. C. imberbe; 6. C. kraussii; 7. C. ne/sonii; 8. C. elaeagnoides; 9. C. hereroense; 10. C. apicu/atum; II. C. panicu/atum; 12. C. microphyllum. A - mollie acid (I); B­ mollie acid P-D-xyloside; C - mollie acid a-L-arabinoside; D - mollie acid P-D-glucoside; E - ( + )-catachin.

and glucoside. However, whereas mollie glucoside is the major constituent and mollie arabinoside the minor constituent in C. molle, the situation is reversed in C. edwardsii, which contains barely discernible quantities of the glucoside but large quantities of the arabinoside and xyloside. This similarity plus the fact that all C. molle specimens examined contained the above four compounds suggests that; (i) these two species may share a common ancestry; (ii) the mollie compounds are extremely ancient chemical characters whose enzyme systems developed before C. molle could develop the different forms mentioned by Exell (1978). Consequently these compounds are excellent markers for C. molle. 4. No distinct or obvious chemical correlation between all the species within a particular taxonomic section was evident, 0 C. molle although similarities such as that between C. paniculatum and • C. imberbe C. microphyllum in section Conniventia and C. edwardsii and • C. hereroense C. molle in section Ciliatipetala do exist. This aspect needs Figure 3 Collection localities for C. molle, C. imberbe, C. hereroense. further investigation, since it is possible that certain chemical D - Darwindale; H - Harare; K - Kadoma; PM - Punda Melia; groupings other than those found in the polar extracts from P - Pretoria; S - Skukuza; HV - Hazyview; M - Malelane; these plants might reflect the morphological differences or L - Louwsberg; SL - St. Lucia estuary; DBN - Durban area similarities more closely. A clearer picture should emerge as (Durban, Stainbank Nature Reserve, Shongweni dam). more of the compounds shown to be present by TLC are isolated and identified. For instance only cycloartane tetra­ cyclic triterpenoids have been isolated from C. molle, C. support the recent proposal by VanWyk (1984) that the afore­ elaeagnoides (Osborne & Pegel 1984; Pegel & Rogers 1985) mentioned taxa should be considered as three distinct species. and C. edwardsii (Rogers 1986), whereas C. imberbe (Rogers 2. C. paniculatum and C. microphyllum Klotzsch 1987) and C. kraussii (Osborne 1982) have yielded only Identical extracts were obtained for these two samples which pentacyclic triterpenoids thus far. These are important dis­ suggests either that they form part of one aggregate species, tinctions whose taxonomic relevance can only be adduced once or that they are subspecies rather than two distinct species. more chemical evidence accumulates. Unfortunately it has not been possible to include C. 5. Finally, exploratory results from four species indicate that platypetalum Welw., the third member of the section Conni­ yields of leaf extract could also be species specific. In each ventia in this particular study. case the percentage yield from leaves collected in the Durban 3. C. molle and C. edwardsii Exell. area was compared with those from leaves collected from the Extracts from these two species are remarkably similar in that same species at Sandton, Transvaal. The results are shown they both contain mollie acid (1) and its arabinoside, xyloside in Table 2. 176 S.-Afr. Tydskr. Plantk., 1987, 53(2)

Table 2 Yields of methanol extracts for identical Acknowledgements species collected from (A) the Durban Area, (8) the We are most grateful to the following for collecting or assisting Transvaal with the collection of leaf samples; Mr Tony Abbott, Um­ 070 yield of tamvuna Nature Reserve; Mrs D. Fourie, B.R.I., Pretoria; Species Location methanol extract the Natal Parks Board; Mr W. Gertenbach, Kruger National Park; Mr Tom Muller, National Herbarium, Zimbabwe; Mr C. celastroides Exell A 16,3 G.R. Nichols, Durban Parks Department and Mr Brian B 16,8 Scherer, Natal Herbarium, Durban. C. obovatum F. Hoffm. A 8,9 We also wish to thank Prof K.H. Pegel, Department of B 8,7 C. microphyl/um Klotzsch A 5,7 Chemistry, University of Natal, Durban whose ideas initiated B 6,2 this work and Mrs L van Hooff, Faculty of Science Photo­ C. kraussii Hochst A 9,3 grapher, for her photographic work. B 8,5 References EXELL, A.W. 1970. Summary of the Combretaceae of Flora Zambesiaca. Kirkia 7: 159-252. EXELL, A.W. 1978. Combretaceae. In: Flora Zambesiaca, ed. Conclusions Launert, E. Vol. 4, pp. 100-183, Flora Zambesiaca Managing The described procedures makes possible the compilation of Committee, London. a register of chemical profiles for species within the genus HARBORNE, J.B., MABRY, T.J. & MABRY, H. 1975. The Combretum that can be used for taxonomic studies and for Flavonoids, Chapman and Hall, London. LOOMIS W.D. 1967. Biosynthesis and metabolism of the rapid identification of species without the necessity of monoterpenes. In: Terpenoids in plants, ed. Pridham J.B. referring to flowers or fruit. It can possibly also be applied Academic Press, London and New York. to the analysis of other families, although individual species OSBORNE, R. 1982. Triterpenoid compounds isolated from leaves may not have the same rich diversity and consistency of polar of Combretum kraussii and C. elaeagnoides. M.Sc. thesis, compounds found in the genus Combretum. In that case other Univ. of Natal. extracts would need to be analysed for chemical characters OSBORNE, R. & PEGEL K.H. 1984. Jessic acid and related acid triterpenoids from Combretum elaeagnoides. Phytochem. 23: that could be used to 'finger-print' the species within the 635 - 637. family or genus. PEGEL, K.H. & ROGERS, C.B. 1985. The characterisation of Because of the sensitivity of TLC analysis, this method mollie acid 3/3-D-xyloside and its genuine aglycone mollie acid, could also be used to analyse herbarium samples, although two novel 1a-hydroxycycioartenoids from Combretum mol/e. J. the effects and extent of decomposition (for example by Chem. Soc. Perkin Trans. I: 1711 - 1715. autoxidation) of chemical constituents in very old samples ROGERS, C.B. & THEVAN I. 1986. Identification of mollie acid a-L-arabinoside, a 1a-hydroxycycioartenoid from Cambre/um would need to be investigated. However, providing samples mol/e leaves. Phytochem. 25: 1759 - 1761. do not differ markedly in age and have been stored under ROGERS, C.B. 1986. Unpublished results. similar conditions, comparative analysis results should be ROGERS, c.B. 1987. The structure of imberbic acid and valid. 23-hydroxy imberbic acid a-L-rhamnoside, two new pentacyciic Since our investigations to date indicate that leaf extracts triterpenoids from Cambretum imberbe. (in press). give species specific chemical profiles, we suggest that when SMITH, P.M. 1976. The chemotaxonomy of plants, Edward Arnold, London. new species are described in the literature, TLC data obtained VAN WYK, A.E. 1984. A new species of Cambretum from from the appropriate leaf extract should be included as one Venda and taxonomic. notes on section Angustimarginata of the diagnostic characters. (Combretaceae). S. Afr. J. Bot. 3: 115 - 119.